Based on the understanding of the ‘way AI works’ and the practice of Kneading the prompts and sources I tried some ‘clean language’ prompts as a way of extracting the themes without bias and remaining curious. It partially addresses the self-tuning of LLMs to the point that they become an echo-chamber.

Of course the idea is not new (How Clean Language Can Help You Become a Better AI Prompt Engineer)but it is important as it establishes guidelines for the Kneading process.